Semiconductors – Supercycle pt. IX

Micron pays the piper.

• Micron is cutting its expectations and spending for 2023 in order to deal with weak demand, but the impact of supply has yet to be felt meaning that this cycle could be as nasty on the way down as it was great on the way up.
• Micron reported FQ1 revenue / EPS of $4.1bn / LOSS$0.04 pretty much in line with forecasts of $4.1bn / LOSS$0.01 which is why the response in after-hours trading was pretty muted.
• However, the company is also making cuts to capex and laying off 10% of staff meaning that the outlook for 2023 remains pretty difficult.
• Capex will now be $7.0bn – $7.5bn which is a leg down from previous expectations and a very large cut from a budget of $12bn discussed less than a year ago.
• The problem is that as a component supplier, Micron is hostage to both end demand and fluctuations in inventory.
• During the pandemic, device makers panicked about supply and raced to increase their inventories for fear of not being able to procure components as a result of the supply problems that the pandemic created.
• This combined with the surge in demand for personal devices from an idle workforce and demand for devices to enable learning and working from home, created a double whammy for component demand.
• Unfortunately, now the reverse is true, and Micron and many others are now having to pay the piper as demand weakening has caused device makers to both reduce inventory as well as sell fewer products.
• This is what has triggered the beginning of the downward turn of the supercycle that I have been fretting about for 15 months (see here).
• As usual, when I get things right, I often get them right for the wrong reason and this case is no different.
• I predicted a surge in capacity would trigger the downturn but instead, it was inflation-triggered demand weakness that caused the cycle to turn.
• The big question now is what impact is capacity going to have.
• Semiconductor fabs are currently being built for geopolitical reasons and not in response to demand and still the output of these factories will have to be absorbed by the market when they come onstream.
• At the same time, I see China turning more towards the lagging edge meaning that supply at 45nm – 28nm may also jump upwards.
• Consequently, the impact of the capex splurge that has been experienced over the last couple of years has yet to be felt, and if it comes during 2023, then the cycle is likely to take another big leg down.
• A lot of this depends on what happens to interest rates and while I don’t think that they will go up much more, I can’t see them coming down either.
• Hence, I am not expecting a recovery in demand next year and expect that 2023 will look similar to 2024.
• By 2024, there is scope for a recovery which is going to be needed as there will be plenty of new fabs by then.
• Against this backdrop, one does not want to be holding the expensive end of the semiconductor sector nor the part of it that owns or is building capacity.
• This is why although TSMC and Samsung look very attractive from a valuation perspective, I prefer Qualcomm and MediaTek, both of whom have a similar valuation but own no factories.